Strange code generated for multiplication #750
Comments
jpellegrini
added a commit
to jpellegrini/STklos
that referenced
this issue
Mar 20, 2025
```
stklos> (disassemble-expr '(* a 2))
000: GLOBAL-REF 0
002: IN-SINT-MUL2 2
004: INT-PUSH 2
006: GLOBAL-REF 0
008: IN-MUL2
```
We're compiling twice!
This is because of a wrongly put parenthesis in `compiler.stk`:
```
((*) (case len
((0) (emit 'IM-ONE))
((1) (if (number? (car actuals))
(compile (car actuals) env epair tail?)
(compile-normal-call fct actuals len env epair #f)))
((2) (let ((a (car actuals))
(b (cadr actuals)))
(if (small-integer-constant? a)
(oper2 'IN-SINT-MUL2 b a))
(comp2 'IN-MUL2))) ;; OOPS! Will be recompiled always!
(else (compile-normal-call fct actuals len env epair #f))))
```
In the above example, (comp2 'IN-MUL2) was supposed to be the alternative for the IF,
but it is outside it, so will always be executed.
This PR fixes it.
Fix egallesio#750
Passes all tests.
Makes the execution time of * becom half what it was before (obviously, but
I measured anyway).
jpellegrini
added a commit
to jpellegrini/STklos
that referenced
this issue
Mar 25, 2025
```
stklos> (disassemble-expr '(* a 2))
000: GLOBAL-REF 0
002: IN-SINT-MUL2 2
004: INT-PUSH 2
006: GLOBAL-REF 0
008: IN-MUL2
```
We're compiling twice!
This is because of a wrongly put parenthesis in `compiler.stk`:
```
((*) (case len
((0) (emit 'IM-ONE))
((1) (if (number? (car actuals))
(compile (car actuals) env epair tail?)
(compile-normal-call fct actuals len env epair #f)))
((2) (let ((a (car actuals))
(b (cadr actuals)))
(if (small-integer-constant? a)
(oper2 'IN-SINT-MUL2 b a))
(comp2 'IN-MUL2))) ;; OOPS! Will be recompiled always!
(else (compile-normal-call fct actuals len env epair #f))))
```
In the above example, (comp2 'IN-MUL2) was supposed to be the alternative for the IF,
but it is outside it, so will always be executed.
This PR fixes it.
Fix egallesio#750
Passes all tests.
Makes the execution time of * become half what it was before (obviously, but
I measured anyway).
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Hi @egallesio .
Look at this:
Good!
Now,
Uh, why? It's doing it twice. Once with the
IN-SINT-MUL2instruction, and then again callingIN-MUL2. The result is correct, but slower than necessary. The operations-,/work fine. Only mutiplication has this issue.The text was updated successfully, but these errors were encountered: